Classification of cutaneous leishmaniasis patients based on its risk factors using latent class analysis

Abstract

Cutaneous leishmaniasis is one of the main health-economic problems around the world. Data were collected from all patients with cutaneous leishmaniasis referred to the health centres of Kermanshah province between 2013 and 2019. Latent class analysis was conducted by PROC LCA in SAS 9.2 and a significant level was set at 0.05. Four latent classes were identified: low (33.8%), moderate (9.8%), high (22.4%) and very high risk (34.0%). The probability of having a travel history was high in the third class. Our study indicated that having history of an eschar has no role in the classification of patients. On the other hand, a positive smear test is important in classifying subjects. Our results indicate that more than half of all patients fell under high risk or very high-risk class. This emphasises the importance of planning preventive intervention by considering different risk factors of cutaneous leishmaniasis simultaneously.

Keywords

Cutaneous leishmaniasis subgrouping latent class analysis Iran

Introduction

Cutaneous leishmaniasis is a preventable zoonosis caused by the parasites of the Leishmania type and transmitted to humans by the bite of different species of the female Phlebotominae sand fly.¹ In Iran, Leishmania tropica (L. tropica) is the cause of anthroponotic cutaneous leishmaniasis and Leishmania major (L. major) is the cause of zoonotic cutaneous leishmaniasis.² The disease is amongst one of the main health-economic problems around the world, including Iran. Currently, more than 350 million people in 101 countries are susceptible to infection, and 1.7 million people are annually infected by cutaneous leishmaniasis, 90% of whom are situated in Afghanistan, Algeria, Brazil, Iran, Iraq, Peru, Saudi Arabia and Syria.^3,4 Despite recent advances in controlling disease, leishmaniasis is still one of the main health issues in some countries such as Iran. Given its significance, the World Health Organization (WHO) has put this disease on the list of the 10 most important diseases of tropical and subtropical regions.⁵

Cutaneous leishmaniasis has existed in Iran for a long time, and Iran is now one of the foci of the disease in the world. The annual recorded number of cutaneous leishmaniasis cases is 20,000 in Iran, whereas the real number of infected cases is more than this number, placing this disease the second most significant parasitic disease after malaria in the country.⁶ Its epidemiology shows that several agents are implicated; sand flies as carriers, wild rodents as reservoirs; several other factors include species of parasite, topography, temperature, relative humidity, precipitation, vegetation, human population and the gender of people.^7,8 The prevalence of the disease globally depends on urbanisation, immigration, population relocation, intercity travel, environmental changes, natural disasters and wars. An outbreak of cutaneous leishmaniasis usually results from the immigration of sensitive people to endemic areas, moving the foci of the disease from one place to another.^9,10 Waves of immigration from neighbouring countries may also accelerate the prevalence of the disease. Although many studies have investigated the risk factors of cutaneous leishmaniasis, the combination of such factors has seldom been studied thus far.

Latent class analysis (LCA) is an approach to identify latent subgroups or classes among participants of a study. This analysis is person-centred and uses some indicator data to identify latent subgroups or sub classes. The underlying basis of the formation of these subgroups is the existing similarities regarding indicator variables.¹¹ The co-existence of cutaneous leishmaniasis risk factors is possible by subgrouping of patients based on such risk factors. To prevent or treat cutaneous leishmaniasis disease properly, it is very important to figure out which risk factors may co-exist. Thus, our study set out to investigate the classification of cutaneous leishmaniasis patients based on risk factors in western Iran.

Materials and methods

Data were collected from all cutaneous leishmaniasis patients referred to the health centres of Kermanshah Province between 2013 and 2019; details were classified based on age, gender, location, the type of ulcer and the species of parasite. LCA was used to analyse the data. Several indicator variables were selected to classify the patients. These were: occupation (five levels), location (two levels), history of travel (two levels), species of parasite in the region (three levels), simultaneous affectation with other diseases (two levels), result of positive smear test (two levels) and having ulcers (two levels). The clustering nature of risk factors was regarded as a latent variable in this study. It is customary to label classes based on item response probabilities >0.5 in the LCA. SPSS 25 was used to report the frequencies of indicator variables. PROC LCA was used in the SAS 9.2 software to conduct the LCA.

Results

A total of 1881 people were infected by cutaneous leishmaniasis over the six-year period of the study with a mean age of 34.8 years and a standard deviation of 15.3 years. Among all of patients 1426 people (75.8%) were male. Table 1 depicts the frequency of the relevant variables of the study. The majority (57.2%) lived in cities and 66.5% had travelled at least once in the past year. The smear test was positive for 79.2% and only 1.7% had ulcers. We attempted to fit the LCA models with classes ranging from 1 to 8. As depicted in Table 2, the relevant indexes were measured for each model. To find the best model, the G² index was first measured (this index is similar to the χ² index). The Akaike information criterion (AIC) and Bayesian information criterion (BIC) indices were measured based on the G² index. Low levels of the three indexes show the acceptability of the model. As can be seen in Table 2, the level of the BIC index was the lowest in the four-class model, and the level of AIC index was the lowest in the five-class model. Given the amounts of such indexes and the interpretability of the result, the four-class model was preferred. Table 3 shows the results of the four-class model for cutaneous leishmaniasis patients.

Table 1.

Distribution frequency of patients according to the studied risk factors by gender.

Items	Male (n = 1426)		Female (n = 455)		Total (n = 1881)
Items	N	%	N	%	N	%
Occupation
Employee	423	29.7	8	1.8	431	22.9
Farmer, rancher, labourer	238	16.7	10	2.2	248	13.2
Driver	427	29.9	0	0	427	22.7
Unemployed	54	3.8	5	1.1	59	3.1
Others	284	19.9	432	94.9	717	38.1
Location
Urban	877	61.5	199	43.7	1076	57.2
Rural	520	36.5	245	53.8	765	40.6
Have a travel history	1060	74.3	191	42.0	1252	66.5
Common parasite type in the area
Leishmania major	344	24.1	163	35.8	507	26.9
Leishmania tropica	150	10.5	59	13.0	209	11.1
Unknown parasite	715	50.1	149	32.7	864	45.9
Simultaneous infection with other diseases	139	9.7	69	15.2	208	11.1
Positive smear test	1143	80.2	348	76.5	1491	79.2
Eschar history	23	1.6	9	2.0	32	1.7

Table 2.

Comparison of LCA models with different latent classes based on model selection statistics.

Number of latent class	Number of parameters estimated	G²	df	AIC	BIC	Maximum log-likelihood
1	11	1389.45	468	1411.45	1472.39	−7512.35
2	23	678.12	456	724.12	851.54	−7156.68
3	35	431.65	444	501.65	695.55	−7033.45
4	47	268.60	432	362.60	622.98	−6951.92
5	59	215.49	420	333.49	660.35	−6925.37
6	71	194.11	408	336.11	729.46	−6914.68
7	83	174.11	396	340.11	799.94	−6904.68
8	95	155.16	384	345.16	871.47	−6895.21

LCA: latent class analysis; AIC: Akaike information criterion; BIC: Bayesian information criterion.

Table 3.

The four latent class model of cutaneous leishmaniasis risk factors.

	Latent class
	Low risk	Moderate risk	High risk	Very high risk
Latent class prevalence	0.39	0.09	0.22	0.34
Item-response probabilities	Probability of a ‘Yes’^a
Occupation
Employee	0.01	0.73	0.54	0.09
Farmer, rancher and worker	0.23	0.05	0.00	0.14
Driver	0.03	0.09	0.02	0.62
Unemployed	0.04	0.09	0.03	0.02
Others	0.68	0.21	0.40	0.12
Location
Urban	0.66	0.99	0.02	0.27
Rural	0.33	0.01	0.97	0.72
Have a travel history	0.28	0.46	0.83	1.00
Common parasite type in the area
Leishmania major	0.48	0.13	0.47	0.11
Leishmania tropica	0.17	0.04	0.29	0.01
Unknown parasite	0.34	0.82	0.23	0.87
Simultaneous infection with other diseases	0.04	0.55	0.14	0.052
Positive smear test	1.00	0.96	0.97	0.99
Eschar history	0.01	0.08	0.09	0.01

^aItem-response probabilities > 0.5 in bold to facilitate interpretation.

The first section of Table 3 shows the prevalence of each of the classes. As depicted, approximately 34% of patients were placed in the first class (low-risk), 10% in the second class (moderate-risk), 22% in the third class (high-risk) and finally 34% in the fourth class (very high-risk). The second section of Table 3 shows the item response probabilities of each risk factor in which probabilities higher than 0.5 are highlighted to emphasise and facilitate interpretation. Accordingly, four latent classes were identified among the patients suffering from cutaneous leishmaniasis with the following features:

The low-risk class had unique features. They were not employees, workers, farmers, drivers or jobless. It is highly possible that these patients were mostly housewives and lived in urban rather than in rural areas. Furthermore, there was a high probability that these patients had positive results for their smear tests. These patients were mostly infected by the L. major parasite.

The moderate-risk class had different features compared to the first class. There was a high probability that they were employed and lived in urban areas. In addition, they were likely to suffer from other types of diseases while infected by cutaneous leishmaniasis.

The high-risk class had very similar features to the second class with the only difference that they mostly lived in rural areas. The probability of having a travel history was also high in this class. Moreover, the prevalent parasite for this class of patients was L. major.

The very high-risk class were mostly drivers, all of whom having a history of travel. Nevertheless, the species of parasite prevalent among these patients was unknown. There was, however, a 100% certainty of a positive smear result in this class.

Thus, the results of our study revealed that history of travel was one of the main, highly probable risk factors of cutaneous leishmaniasis in the first and fourth classes. Besides, positive smear results also had a very high probability (near 100%) in all classes. Yet, the ulcer factor did not have any significant role in the classification of the patients. In terms of occupation, it was observed that neither farming, being a rancher or worker gave a high probability in any classes.

Discussion

The reason why men are more infected than women may be because of the type of clothes worn, as men usually wear lighter and thinner clothes compared to women. Another reason could be the fact that men travel or migrate more than women and hence are more exposed to the bite of sand flies in endemic regions.

Previous studies have used the LCA model to evaluate the sensitivity and specificity of different tests for the diagnosis of leishmaniasis.¹² The main advantage of this method over others is that it can show the co-existence possibility of risk factors. Kermanshah Province shares a border with the neighbouring country, Iraq; many people travel annually to Iraq for pilgrimage. Drivers are highly susceptible to infection because of close contact with their passengers carrying the disease across the border. Several other studies conducted in different areas in Iran, such as Hamedan and Damghan, have shown that travel in endemic areas is one of the key risk factors.^13,14

Moreover, we found that the most prevalent species of parasite was L. major. Therefore, the distribution of the disease in the study region is arguably broader in rural than in urban areas. Cutaneous leishmaniasis is more prevalent in rural regions in Iran and more than 17 provinces suffer from the disease.^15,16 Reasons for this wide distribution is the great diversity of hosts of the leishmaniasis parasite include the many species of wild rodents and the wide diversity of sand flies.^17,18

It is therefore highly recommended that passengers and especially drivers be educated about personal hygiene in order to implement appropriate measures for its prevention.

Footnotes

Authors’ contributions

MS, EMA, AAGH and ESH designed the work and edited the article. MS and KHK collected the data. EM and AAGH performed the data analysis. ESH and MS performed critical revisions of the article.

Acknowledgements

The authors are grateful to all colleagues at the Kermanshah University of Medical Sciences and the staff at the health centres in the all counties in Kermanshah Province.

Declaration of conflicting interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

ORCID iD

Eslam Moradi-Asl

References

Farahmand

Nahrevanian

Shirazi

, et al. An overview of a diagnostic and epidemiologic reappraisal of cutaneous leishmaniasis in Iran. Braz J Infect Dis 2011; 15: 17–21.

Reithinger

Dujardin

J-C

Louzir

, et al. Cutaneous leishmaniasis. Lancet Infect Dis 2007; 7: 581–596.

Postigo

JAR

. Leishmaniasis in the World Health Organization Eastern Mediterranean region. Int J Antimicrob Agents 2010; 36: S62–S65.

Organization WHO. Report on the interregional meeting on leishmaniasis among neighbouring endemic countries in the Eastern Mediterranean, African and European regions, Amman, Jordan, 23–25 September 2018. World Health Organization, Regional Office for the Eastern Mediterranean, 2019.

Organization WHO. Working to overcome the global impact of neglected tropical diseases: first WHO report on neglected tropical diseases. Geneva: World Health Organization, 2010.

Mohebali

. Visceral leishmaniasis in Iran: review of the epidemiological and clinical features. Iran J Parasitol 2013; 8: 348–358.

Mohebali

Moradi-Asl

Rassi

. Geographic distribution and spatial analysis of Leishmania infantum infection in domestic and wild animal reservoir hosts of zoonotic visceral leishmaniasis in Iran: a systematic review. J Vector Borne Dis 2018; 55: 173–173.

Moradi-Asl

Adham

Hanafi-Bojd

, et al. Spatial distribution of sand flies (Diptera: Psychodidae; Larroussius group), the vectors of visceral leishmaniasis in Northwest of Iran. Asian Pac J Trop Biomed 2018; 8: 425–430.

Pratlong

Dereure

Ravel

, et al. Geographical distribution and epidemiological features of Old World cutaneous leishmaniasis foci, based on the isoenzyme analysis of 1048 strains. J Trop Pediatr 2009; 14: 1071–1085.

10.

Moradi-Asl

Hanafi-Bojd

Rassi

, et al. Situational analysis of visceral leishmaniasis in the most important endemic area of the disease in Iran. Iran J Arthropod Borne Dis 2017; 11: 482–482.

11.

Abbasi-Ghahramanloo

Janani

Malakouti

, et al. Risk taking behaviors in relation to Attention Deficit and Hyperactivity Disorder in Iranian male workers: a latent class analysis. Neuropsychiatr Dis Treat 2019; 15: 2513–2513.

12.

da Silva Solca M, Bastos LA, Guedes CES, et al. Evaluating the accuracy of molecular diagnostic testing for canine visceral leishmaniasis using latent class analysis. PloS One 2014; 9.

13.

Rafati

Shaporimoghadem

Ghorbani

. Epidemiological study of cutaneous Leishmaniasis in Damghan (2000-2006). Koomesh 2007; 8: 247–254.

14.

Zahirnia

Moradi

Norozi

, et al. Epidemiological survey of cutaneous Leishmaniasis in Hamadan province (2002-2007). Avicenna J Med Biotechnol 2009; 16: 43–47.

15.

Rasi

Javadian

Jalali

, et al. Investigation on zoonotic cutaneous leishmaniasis, Southern Iran. Iran J Public Health 2004; 33: 31–35.

16.

Yaghoobi-Ershadi

Javadian

. Epidemiological study of reservoir hosts in an endemic area of zoonotic cutaneous leishmaniasis in Iran. Bull World Health Organ 1996; 74: 587–587.

17.

Gholamrezaei

Mohebali

Hanafi-Bojd

, et al. Ecological niche modeling of main reservoir hosts of zoonotic cutaneous leishmaniasis in Iran. Acta Trop 2016; 160: 44–52.

18.

Yaghoobi-Ershadi

. Phlebotomine sand flies (Diptera: Psychodidae) in Iran and their role on Leishmania transmission. Iran J Arthropod Borne Dis 2012; 6: 1–10.